Alaskans will vote on a ranked-choice voting system this November

[jimrtex]

Passage/failure is going to be really close (within 1%).

Alaska rarely has 4 candidates running for legislative seats. The exceptions are open Republican seats, where multiple candidates will run knowing that if they are successful, they will almost certainly win the general election. But in seats like that, there may not be even one Democrat candidate. The Democrats have even got an interpretation of the state constitution that they can nominate independents. There has been some back and forth as to how they appear on the ballot.

Alyse Galvin wanted to be listed as

(    )  Alyse Galvin   Non-Partisan

or at least:

(    ) Alyse Galvin   (N)     Democratic nominee

but courts ruled that

(    ) Alyse Galvin          Democratic nominee

accurately represented the basis for her appearance on the ballot. In 2018, the second style had been used, though Galvin was Undeclared or (U) at the time

Non-partisan and Undeclared are a majority in Alaska at 56.8%, with Republicans 24.6%, and Democrats 13.6%.

Alaska had used a blanket primary since territorial days, having borrowed it from Washington which had used it since the 1930s. In a blanket primary, a voter may choose which primary to vote in for each office, helping choose the Republican nominee for Governor, and the Democratic nominee for Senator, etc.

After California adopted the blanket primary in the late 1990s, the SCOTUS ruled it unconstitutional claiming it permitted Republicans to interfere in Democratic nominations etc. California and Washington switched to conventional partisan nominations (though in Washington affiliation was secret).

Before this was going to happen, parties in Alaska got a ruling from the Alaska Supreme Court that the Alaska Constitution affords greater protection to freedom of association than the 1st Amendment of the US Constitution does. A party may invite voters to participate in its nomination contests, while the State may not compel them to do so.

All parties but the Republicans permitted all voters to participate in their primaries. So a Republican voter could take a composite ballot and vote for the Democratic nominee for governor, Alaskan Independence nominee for lieutenant governor, Libertarian for senator, etc.

Washington and California have since adopted Top 2, and if Alaska adopts Top 4, all three blanket primary states will have adopted some form of non-partisan nomination.

Alaska all but encourages write-in campaigns, including by losers of primaries. When Lisa Murkowski lost the Republican primary, a casual reading of the statute on write-ins could be interpreted as
"Run Lisa Run!"

Under Top 4 this could mean that most legislative races have write-in candidates qualifying for the general election. In Washington and California, write-in candidates can qualify but finishing in the Top 2 is much harder than finishing in the Top 4. It usually only happens when there is an unopposed major candidate. If it is hopeless for a Republican to run, it is even more hopeless for a Libertarian or Green to run.

Alaska legislative districts are small (in terms of voters) so it is feasible to campaign door-to-door in at least in cities, and they appear to have elected two real independents.

While it will retain partisan nominations, the presidential general election will be by RCV. There were 7 presidential on-ballot candidates, and the Alaskan Independence Party did not have a candidate.

[jimrtex]

Passage/failure is going to be really close (within 1%).

My models say it should pass by 1.5 to 2 points. We'll see.

Yes closed to within 2,000 votes after tonight's early count dump, but a lot of them were from D-leaning HDs.

There's a fairly strong correllation - r^2 of 0.89 or so - between the D-R registration advantage and the absentees reported by HD so far. So I have a number of ways to calculate this. As I said, all ways point to a RCV win here.

Today's dump was the strongest yet for Yes, but it included HD-17 and HD-18 in central Anchorage, HD-32 Miscellaneous, Et Cetera, and Other in Southern Alaska (that is a really weird district - what are the population splits?)

Are the only districts still out HD-13, HD-22, HD-36, HD-37, HD-38, and HD-39, or are there some incomplete?

I'm guessing that HD-13, HD-22, and HD-36 will have about 3000 votes each with 60% Yes which will about close the gap. How many mail ballots in HD-37, HD-38, and HD-39. If the 633 from HD-40 is complete, then not too many. In all the tiny villages it is a lot easier to vote in person than to mail a ballot in - the polling place may have longer hours than the post office does, and they may not have residential delivery.

I'm still inclined to think it will be under 1%.

[jimrtex]

Passage/failure is going to be really close (within 1%).

My models say it should pass by 1.5 to 2 points. We'll see.

Yes closed to within 2,000 votes after tonight's early count dump, but a lot of them were from D-leaning HDs.

There's a fairly strong correllation - r^2 of 0.89 or so - between the D-R registration advantage and the absentees reported by HD so far. So I have a number of ways to calculate this. As I said, all ways point to a RCV win here.

Today's dump was the strongest yet for Yes, but it included HD-17 and HD-18 in central Anchorage, HD-32 Miscellaneous, Et Cetera, and Other in Southern Alaska (that is a really weird district - what are the population splits?)

Are the only districts still out HD-13, HD-22, HD-36, HD-37, HD-38, and HD-39, or are there some incomplete?

I'm guessing that HD-13, HD-22, and HD-36 will have about 3000 votes each with 60% Yes which will about close the gap. How many mail ballots in HD-37, HD-38, and HD-39. If the 633 from HD-40 is complete, then not too many. In all the tiny villages it is a lot easier to vote in person than to mail a ballot in - the polling place may have longer hours than the post office does, and they may not have residential delivery.

I'm still inclined to think it will be under 1%.

Here's my estimate of what ABSENTEES are out as of right now:



40 is not complete. Some of it is counted by the Nome office; some of it is counted by Fairbanks. Aleutians 37 has 1,519; Bethel 38 has 1,135; Nome 39 has 592 and Barrow/Kotzebue 40 has 400 (probably in parts closer to Fairbanks than Nome).

The big district yet to report is 31 on the Southern Kenai Peninsula. Its absentees are poised to go 57% yes according to my model - which was off on 32. 13's absentees are expected to go 55% Yes.

So far, the absentees have only gone No in uber-Republican HDs 8 and 30.

There also are some early and Questioned/provisional votes to be counted, but those have been more or less been breaking even so far.

Was there a release of number of absentee ballots?

After the second dump on Thursday included HD-13, 22, 31; that left HD-36, 37, 38, and 39 as the only districts with no absentee ballots reported,

So were the remaining ballots the last-arriving or why weren't they counted?

For example, HD-24 had 2435 votes counted in one big batch, but you show 504 more ballots to be counted.

[jimrtex]

No sooner do I post that then another 11.5K or so votes dropped, and BM 2 took the lead.

Here's the revised estimated what's left charts. The numbers are rough, so put a little bit of a range on it.

Why are the numbers different for districts like HD-29 (+200), HD-37 (+51), HD-14 (+54), HD-38 (+13).

I was looking at the roll-off to determine if there was any difference based on mode of voting. Based on a small sample (HD-15), it appears not. (4.8% vs 4.9%). Once upon a time, it was conventional wisdom to vote against any proposition you didn't understand. But I think over time that may have changed. Those more susceptible to persuasion might vote Yes to go along with the crowd ("If I vote Yes, others will like me").

There was likely XX% of voters who had no idea that there would be any referendums on the ballot, let alone what they were about, this is true of everything but President ("What is this "Senator" about. I wonder if this Don Young is new. I've never heard of him.").

I thought it possible that those who voted in-person might have a different response than those who voted by mail. If you were filling out a ballot at home, you might be more likely to research questions. It would be like a take-home or open-book test.

But this does not seemed to be the case.

But what I noticed is the roll off for questioned ballots was extreme. Swarms were not getting past President, and most weren't voting for the legislature let alone the questions.

CUCUKLILLRUUNGA!

(This is Yup'k for "I voted")

[jimrtex]

Was there a release of number of absentee ballots?

After the second dump on Thursday included HD-13, 22, 31; that left HD-36, 37, 38, and 39 as the only districts with no absentee ballots reported,

So were the remaining ballots the last-arriving or why weren't they counted?

For example, HD-24 had 2435 votes counted in one big batch, but you show 504 more ballots to be counted.

They're counting in batches, up to a certain date received. The remaining absentee ballots outside of 36-39 and maybe 29 are probably late arrivals.

There was only one Friday dump, which was the weakest Yes vote yes. But that is likely because if was mostly from the Fairbanks and Mat-Su districts. In every district, the Yes percentage increased, 0.1% - 0.3%. But the new votes were just a little. Perhaps they were just bumping the certain date.

The regional offices are behaving differently. Fairbanks already had the least remaining ballots, but did s few more. Mat-Su districts had relatively few remaining but did some more.

It appears Anchorage mostly took the day off. HD-18 may just have been finishing off what they had started on Thursday. Otherwise they only added 146 votes in HD-22.

Juneau did the first dump in HD-36, and a lot more in HD-31, which simply has bunches of mail ballots.

Nome appears to be holding out. Will they be done by Christmas?

[jimrtex]

The Alaska Division of Elections put out a formal list of what’s been counted and what’s still out last night:
http://www.elections.alaska.gov/results/20GENR/data/sovc/unofficialballotcountstats.pdf

Thanks!

Looking at the 16 Anchorage districts (HD-13 to HD-28), if the remaining absentee votes have the same distribution as those counted, that would add +2962 to the margin, increasing it from 0.36% to 0.94%. But I did not account for the undervote. Let's say 5%, which reduce the gain to about 2800, and the new margin to 0.88% (this again is just for Anchorage).

It appears that about 13% to 22% of the absentee mail ballots have not been counted. Since you said they were taking into account the date of arrival, this might be those that arrived on or after election day, or at least October 31. These may be more Republican/No than the earlier arrivals, just like is happening in California. Because of the liberal return deadline in Alaska, voters might think nothing of putting their ballot in the mail on election day or perhaps on Monday. They will have no concern that it might take 5 days to go across Anchorage.

The lowest share of uncounted mail ballots is 13% in HD-28 in the southern Anchorage including some quite rural areas. There may be some areas without home mail delivery. If you have a mailbox on the side of the road, are you going to stick anything of value like a ballot in it? By now you may have pulled it out of the ground, and transferred all important mail to to a postal box.

Other than HD-28, the districts with the highest shares of uncounted absentees ae HD-13, 14, and 15 which gives a possibility that these late arrivals are more Republican/No.

But they would have to maybe 14% more No to wipe out the net positive Yes for the last uncounted Anchorage mail ballots. This seems like a stretch.

While I was writing this, the mid-day Saturday dump completed the results from the Juneau regional center. I'll see if I can differentiate (I had only save an overall percentage, so there is going to be some rounding problems).

[jimrtex]

The Alaska Division of Elections put out a formal list of what’s been counted and what’s still out last night:
http://www.elections.alaska.gov/results/20GENR/data/sovc/unofficialballotcountstats.pdf

Thanks!

Looking at the 16 Anchorage districts (HD-13 to HD-28), if the remaining absentee votes have the same distribution as those counted, that would add +2962 to the margin, increasing it from 0.36% to 0.94%. But I did not account for the undervote. Let's say 5%, which reduce the gain to about 2800, and the new margin to 0.88% (this again is just for Anchorage).

It appears that about 13% to 22% of the absentee mail ballots have not been counted. Since you said they were taking into account the date of arrival, this might be those that arrived on or after election day, or at least October 31. These may be more Republican/No than the earlier arrivals, just like is happening in California. Because of the liberal return deadline in Alaska, voters might think nothing of putting their ballot in the mail on election day or perhaps on Monday. They will have no concern that it might take 5 days to go across Anchorage.

The lowest share of uncounted mail ballots is 13% in HD-28 in the southern Anchorage including some quite rural areas. There may be some areas without home mail delivery. If you have a mailbox on the side of the road, are you going to stick anything of value like a ballot in it? By now you may have pulled it out of the ground, and transferred all important mail to to a postal box.

Other than HD-28, the districts with the highest shares of uncounted absentees ae HD-13, 14, and 15 which gives a possibility that these late arrivals are more Republican/No.

But they would have to maybe 14% more No to wipe out the net positive Yes for the last uncounted Anchorage mail ballots. This seems like a stretch.

While I was writing this, the mid-day Saturday dump completed the results from the Juneau regional center. I'll see if I can differentiate (I had only save an overall percentage, so there is going to be some rounding problems).

Based on the final results from Juneau we can calculate the Yes percentage from the mail  ballots counted on Saturday, compared to that for mail ballots counted previously, and all non-mail ballots (election day, early voting*, and questioned ballots).

I calculated this based on the overall Yes % before Saturday, and the number of mail ballots added Saturday. All calculations are based on the results on the question, and totally exclude any undervotes - effectively, they just never happened. Thus my calculations will differ slightly from if I had recorded the number of Yes and No votes by mail prior to Saturday, but it algebraically correct:

HD	New Mail	Prev Mail	Non-Mail
29	42.8%	49.9%	29.7%
30	41.7%	46.3%	30.0%
31	57.6%	58.6%	35.6%
32	59.6%	66.8%	51.5%
33	70.0%	79.0%	64.3%
34	61.4%	71.3%	53.5%
35	52.6%	67.1%	49.2%
36	64.7%	61.3%	45.4%

The newly counted (later arriving) mail ballots were 5% to 15% less Yes than the earlier-arriving mail ballots. This pattern holds whether the Yes support was high (HD-33 or HD-34), or low (HD-29, 39, and 31).

The exceptions to this are HD-31 and HD-36, but it appears that after finishing the earlier mail ballots on Friday they continued counting the later arriving ballots, leaving only a smidgen to be finished on Saturday. Since they finished counting in mid-day it is possible that signatures had already been validated, and they only had to run the ballots through scanners (for all the Juneau HD),

If we assume that the later arriving Anchorage mail ballots are 7.5% less Yes than the earlier arriving mail ballots (+15% reduction in the margin), then that would reduce the added margin in Anchorage from 2800 to 1300.

*The early voting from the two Juneau districts (HD-33 and HD-34) was huge because it was convenient to get to the regional offices, so a large share of the non-mail ballots in these districts was like the mail ballots prior to election day.

[jimrtex]

Alaska Voters = Cowards for adopting RCV!

How so?

[jimrtex]

The Alaska Division of Elections put out a formal list of what’s been counted and what’s still out last night:
http://www.elections.alaska.gov/results/20GENR/data/sovc/unofficialballotcountstats.pdf

Thanks!

Looking at the 16 Anchorage districts (HD-13 to HD-28), if the remaining absentee votes have the same distribution as those counted, that would add +2962 to the margin, increasing it from 0.36% to 0.94%. But I did not account for the undervote. Let's say 5%, which reduce the gain to about 2800, and the new margin to 0.88% (this again is just for Anchorage).

It appears that about 13% to 22% of the absentee mail ballots have not been counted. Since you said they were taking into account the date of arrival, this might be those that arrived on or after election day, or at least October 31. These may be more Republican/No than the earlier arrivals, just like is happening in California. Because of the liberal return deadline in Alaska, voters might think nothing of putting their ballot in the mail on election day or perhaps on Monday. They will have no concern that it might take 5 days to go across Anchorage.

The lowest share of uncounted mail ballots is 13% in HD-28 in the southern Anchorage including some quite rural areas. There may be some areas without home mail delivery. If you have a mailbox on the side of the road, are you going to stick anything of value like a ballot in it? By now you may have pulled it out of the ground, and transferred all important mail to to a postal box.

Other than HD-28, the districts with the highest shares of uncounted absentees ae HD-13, 14, and 15 which gives a possibility that these late arrivals are more Republican/No.

But they would have to maybe 14% more No to wipe out the net positive Yes for the last uncounted Anchorage mail ballots. This seems like a stretch.

While I was writing this, the mid-day Saturday dump completed the results from the Juneau regional center. I'll see if I can differentiate (I had only save an overall percentage, so there is going to be some rounding problems).

Based on the final results from Juneau we can calculate the Yes percentage from the mail  ballots counted on Saturday, compared to that for mail ballots counted previously, and all non-mail ballots (election day, early voting*, and questioned ballots).

I calculated this based on the overall Yes % before Saturday, and the number of mail ballots added Saturday. All calculations are based on the results on the question, and totally exclude any undervotes - effectively, they just never happened. Thus my calculations will differ slightly from if I had recorded the number of Yes and No votes by mail prior to Saturday, but it algebraically correct:

HD	New Mail	Prev Mail	Non-Mail
29	42.8%	49.9%	29.7%
30	41.7%	46.3%	30.0%
31	57.6%	58.6%	35.6%
32	59.6%	66.8%	51.5%
33	70.0%	79.0%	64.3%
34	61.4%	71.3%	53.5%
35	52.6%	67.1%	49.2%
36	64.7%	61.3%	45.4%

The newly counted (later arriving) mail ballots were 5% to 15% less Yes than the earlier-arriving mail ballots. This pattern holds whether the Yes support was high (HD-33 or HD-34), or low (HD-29, 39, and 31).

The exceptions to this are HD-31 and HD-36, but it appears that after finishing the earlier mail ballots on Friday they continued counting the later arriving ballots, leaving only a smidgen to be finished on Saturday. Since they finished counting in mid-day it is possible that signatures had already been validated, and they only had to run the ballots through scanners (for all the Juneau HD),

If we assume that the later arriving Anchorage mail ballots are 7.5% less Yes than the earlier arriving mail ballots (+15% reduction in the margin), then that would reduce the added margin in Anchorage from 2800 to 1300.

*The early voting from the two Juneau districts (HD-33 and HD-34) was huge because it was convenient to get to the regional offices, so a large share of the non-mail ballots in these districts was like the mail ballots prior to election day.

33 and 34 always lead the early voting. 33 includes some towns nowhere near Juneau in addition to Downtown Juneau. The early votes likely come from heavily R downtown Juneau, not those villages, which are more likely to vote absentee, in person or otherwise.

Your analysis is missing the heavily D, rural AK HDs 37-39, which have a lot of ballots out that likely skew Yes.

I was noting the early voting in HD-33 and HD-34 because of its effect on the "Non-Absentee" vote. In other districts the "Non-Absentee" vote essentially the election day vote plus a tiny number of early votes and question votes.

The early vote in HD-34 was 145 times as great as the early vote in HD-30.

In HD-34

Election Day: 22.3% of votes, 58.5% Yes
Early Voting: 38.9% of votes, 71.1% Yes.
Mail Voting: 35.6% of votes, 77.5% Yes.

Early voting likely cannibalizes from mail voting and election day voting.

I had not/have not completed my analysis.

I did the Anchorage analysis first because it had a lot of districts, lots of mail ballots counted, but also lots of mail ballots to be counted.

I then did Juneau because it was completed on Saturday and it gave a means to differentiate between earlier arrived mail ballots and later arrived mail ballots.

I have completed the analysis of the Fairbanks districts (HD-1 through HD-6, and HD-9). The share of outstanding mail ballots there is smaller than in Anchorage.

If we assume the later arriving mail ballots are as favorable as the earlier arriving mail ballots, then this would produce about a +200 margin for Yes. If they are 5% less favorable than earlier mail ballots, that trims the margin to +125. There are only 750 outstanding ballots in Fairbanks.

I haven't yet looked at the Sunday results. Given that they were only modestly favorable, and were released mid-day, I'm assuming they are either from Fairbanks and/or Mat-Su.

[jimrtex]

Your analysis is missing the heavily D, rural AK HDs 37-39, which have a lot of ballots out that likely skew Yes.

The mid-day Sunday results are from Anchorage, but they didn't update the district tables, I had to go hunting through the "precinct results".

BTW, any idea why there are two sets of absentee results for HD-6 and HD-9?

[jimrtex]

I haven't yet looked at the Sunday results. Given that they were only modestly favorable, and were released mid-day, I'm assuming they are either from Fairbanks and/or Mat-Su.

No. They were from Anchorage. Yes gained about 1,000 votes out of 7.6K, and is now up by 2,580 votes.

It's over. The bush HDs will put it even more up than it already is.

I'm sure it will pass, but it will be interesting to see the margin. It still isn't to 1%.

The new Anchorage mail ballots are about 5% less Yes than earlier ballots. They also aren't finishing up all the ballots.

[jimrtex]

I haven't yet looked at the Sunday results. Given that they were only modestly favorable, and were released mid-day, I'm assuming they are either from Fairbanks and/or Mat-Su.

No. They were from Anchorage. Yes gained about 1,000 votes out of 7.6K, and is now up by 2,580 votes.

It's over. The bush HDs will put it even more up than it already is.

I'm sure it will pass, but it will be interesting to see the margin. It still isn't to 1%.

The new Anchorage mail ballots are about 5% less Yes than earlier ballots. They also aren't finishing up all the ballots.

HD-38 and HD-39  were not particularly rich sources of Yes votes. The mail votes were just a bit more Yes than election day, and there weren't that many mail votes.

In a bush village, mail might not be all that convenient. How often does a plane get in and out with mail?

[jimrtex]

Your analysis is missing the heavily D, rural AK HDs 37-39, which have a lot of ballots out that likely skew Yes.

The mid-day Sunday results are from Anchorage, but they didn't update the district tables, I had to go hunting through the "precinct results".

BTW, any idea why there are two sets of absentee results for HD-6 and HD-9?

I think they're counted or at least collected by different offices - Faribanks vs Nome for 6 and Mat-Su vs Fairbanks for 9. There also are double entries for other HDs in the precinct text files, too, likely for the same reason. That messed up my results spreadsheet, since they changed it on the fly out of the blue in one update. I think it also messed up the NYT and CNN's results, too, at one point. They caught it way after I did.

I've caught up with the Anchorage results from Sunday.

They added 7785 votes, and have around 2250 remaining.

The newer mail ballots were substantially less favorable than earlier mail ballots, while still more favorable than the election day votes. For HD-16 through HD-28

56.6% latest mail ballots.
64.5% earlier mail ballots
47.2% non-mail (election day, questionable, early vote)

The Sunday mail ballots from Anchorage added 1021 to the margin. If the latest mail rate holds for the remaining mail ballots that would add +306.

Starting with current margin of +3218
+306 from Anchorage
+200 from Fairbanks (assumed same mail rate, so perhaps high a bit)
+300 from Mat-Su (just a quick eyeball, more ballots outstanding than Fairbanks)
+636 assuming similar results for HD-37 and HD-40 as HD-38 and HD-39. HD-37 has more ballots, but HD-40 fewer.
+200 for questionable and early votes for HD-37 through HD-40, assuming 60:40 split.
+0 for "HD-99" since they only vote on federal offices.

So a final margin of +4860 which would 1.41% of 345,000 total votes.

Alternative, NO would need 71.5% of remaining ballots to flip result. They didn't quite get that in HD-29 one of very strongest districts.

Has Kenai ever been an island?

[jimrtex]

Happy to hear that this looks likely to pass, but how likely is a bill to implement this to get through the state legislature and be signed by Dunleavy?

It is statute. It will become effective 90 days after it is certified, and is not subject to veto. It may not be repealed for two years after the effective date, though it can be amended. I do not if there are judicial interpretations of how extensive an amendment may be. Since the right of the initiative is in the constitution there may be limitations on neutering an initiative without full repeal.

For example, the Alaska constitution permits the governor to fill legislative vacancies, but reserves to the legislature the authority to provide for filling vacancies. Current statute does not provide for special elections except for senatorial vacancies that occur in the first two years of a 4-year term. The special election is held concurrently with the next general election. Current statute also provides that legislators of the same party and house of the predecessor must confirm the successor. Thus a governor would probably be blocked from appointing a stealth Republican or Democrat. The initiative amended these provisions to align them with the language of the initiative - since there are no longer party nominees, a legislator is a member of a party, rather than the nominee of a party.

The legislature could conceivably amend the statute to provide an RCV special election to fill legislative vacancies, since that would not really be contrary to the spirit or intent of the statute.

Congressional vacancies are currently filled by special election, and senatorial vacancies are filled sooner rather than later. Under the current law, there are no partisan primaries, but there is a contingent runoff if no candidate gets a majority (i.e. the system used in Georgia). The initiative modified these provisions so now the special election is conducted by RCV. So there are already provisions to borrow if the legislature wished to implement special elections for the legislature.

[jimrtex]

Your analysis is missing the heavily D, rural AK HDs 37-39, which have a lot of ballots out that likely skew Yes.

The mid-day Sunday results are from Anchorage, but they didn't update the district tables, I had to go hunting through the "precinct results".

BTW, any idea why there are two sets of absentee results for HD-6 and HD-9?

I think they're counted or at least collected by different offices - Faribanks vs Nome for 6 and Mat-Su vs Fairbanks for 9. There also are double entries for other HDs in the precinct text files, too, likely for the same reason. That messed up my results spreadsheet, since they changed it on the fly out of the blue in one update. I think it also messed up the NYT and CNN's results, too, at one point. They caught it way after I did.

I've caught up with the Anchorage results from Sunday.

They added 7785 votes, and have around 2250 remaining.

The newer mail ballots were substantially less favorable than earlier mail ballots, while still more favorable than the election day votes. For HD-16 through HD-28

56.6% latest mail ballots.
64.5% earlier mail ballots
47.2% non-mail (election day, questionable, early vote)

The Sunday mail ballots from Anchorage added 1021 to the margin. If the latest mail rate holds for the remaining mail ballots that would add +306.

Starting with current margin of +3218
+306 from Anchorage
+200 from Fairbanks (assumed same mail rate, so perhaps high a bit)
+300 from Mat-Su (just a quick eyeball, more ballots outstanding than Fairbanks)
+636 assuming similar results for HD-37 and HD-40 as HD-38 and HD-39. HD-37 has more ballots, but HD-40 fewer.
+200 for questionable and early votes for HD-37 through HD-40, assuming 60:40 split.
+0 for "HD-99" since they only vote on federal offices.

So a final margin of +4860 which would 1.41% of 345,000 total votes.

Alternative, NO would need 71.5% of remaining ballots to flip result. They didn't quite get that in HD-29 one of very strongest districts.

Has Kenai ever been an island?

The remainder of the Mat-Su mail ballots were extremely weak, around 15% less than previous mail ballots, and about the same as the election day ballots.

The net Yes margin was -244. This suggests that the remainder of the Fairbanks and Anchorage mail ballots will not be as positive as the previous mail ballots. They will still be positive, because these areas were stronger Yes than Mat-Su, but it will still cut into the expected margins.

The Yes margin for the Early Vote and Questioned votes in HD-37, HD-38, and HD-39 was +153. These votes were numerically dominated by the Early Vote from Nome in HD-39. Otherwise these votes would have been equivocal.

The current margin is +3027.

Cutting the estimates for Fairbanks and Anchorage in half:

+100 and +150

+636 for mail ballots from HD-37 and HD-40 to match results for HD-38 and HD-39.

+0 for equivocal results for Early Vote and Questionable for HD-40.

Final margin +3900 of about 1.13% (50.6% Yes to 49.4%). This margin is less than the undervote.

[jimrtex]

Your analysis is missing the heavily D, rural AK HDs 37-39, which have a lot of ballots out that likely skew Yes.

The mid-day Sunday results are from Anchorage, but they didn't update the district tables, I had to go hunting through the "precinct results".

BTW, any idea why there are two sets of absentee results for HD-6 and HD-9?

I think they're counted or at least collected by different offices - Faribanks vs Nome for 6 and Mat-Su vs Fairbanks for 9. There also are double entries for other HDs in the precinct text files, too, likely for the same reason. That messed up my results spreadsheet, since they changed it on the fly out of the blue in one update. I think it also messed up the NYT and CNN's results, too, at one point. They caught it way after I did.

I've caught up with the Anchorage results from Sunday.

They added 7785 votes, and have around 2250 remaining.

The newer mail ballots were substantially less favorable than earlier mail ballots, while still more favorable than the election day votes. For HD-16 through HD-28

56.6% latest mail ballots.
64.5% earlier mail ballots
47.2% non-mail (election day, questionable, early vote)

The Sunday mail ballots from Anchorage added 1021 to the margin. If the latest mail rate holds for the remaining mail ballots that would add +306.

Starting with current margin of +3218
+306 from Anchorage
+200 from Fairbanks (assumed same mail rate, so perhaps high a bit)
+300 from Mat-Su (just a quick eyeball, more ballots outstanding than Fairbanks)
+636 assuming similar results for HD-37 and HD-40 as HD-38 and HD-39. HD-37 has more ballots, but HD-40 fewer.
+200 for questionable and early votes for HD-37 through HD-40, assuming 60:40 split.
+0 for "HD-99" since they only vote on federal offices.

So a final margin of +4860 which would 1.41% of 345,000 total votes.

Alternative, NO would need 71.5% of remaining ballots to flip result. They didn't quite get that in HD-29 one of very strongest districts.

Has Kenai ever been an island?

The remainder of the Mat-Su mail ballots were extremely weak, around 15% less than previous mail ballots, and about the same as the election day ballots.

The net Yes margin was -244. This suggests that the remainder of the Fairbanks and Anchorage mail ballots will not be as positive as the previous mail ballots. They will still be positive, because these areas were stronger Yes than Mat-Su, but it will still cut into the expected margins.

The Yes margin for the Early Vote and Questioned votes in HD-37, HD-38, and HD-39 was +153. These votes were numerically dominated by the Early Vote from Nome in HD-39. Otherwise these votes would have been equivocal.

The current margin is +3027.

Cutting the estimates for Fairbanks and Anchorage in half:

+100 and +150

+636 for mail ballots from HD-37 and HD-40 to match results for HD-38 and HD-39.

+0 for equivocal results for Early Vote and Questionable for HD-40.

Final margin +3900 of about 1.13% (50.6% Yes to 49.4%). This margin is less than the undervote.

Anchorage absentee complete +309.
HD-37 absentee complete +91
HD-40 absentee complete +304

There are about 1000 Juneau mail ballots uncounted, and an uncertain number of questioned and early votes in HD-40.

Current margin +3856. Juneau will add +100. Final margin 1.12% (50.56% to 49.44%)

[jimrtex]

Anchorage absentee complete +309.
HD-37 absentee complete +91
HD-40 absentee complete +304

There are about 1000 Juneau mail ballots uncounted, and an uncertain number of questioned and early votes in HD-40.

Current margin +3856. Juneau will add +100. Final margin 1.12% (50.56% to 49.44%)

I don't think it's Juneau that's left. I think it's Fairbanks region's 1,2,4-6 and 9, plus maybe a little bit of 37. 40 should be done counting.

My guess is 830 votes or so remain - but it's just a guess. Could be up to 1,000.

Yes, I meant Fairbanks.

Their absentee ballots were completed on Wednesday, and I think the early vote and questioned vote were completed as well.

The link to the number of outstanding ballots has also disappeared so I think they are done.

Final margin is 1.10%.

An oddity is that the final ballots were heavily undervoted, except for President. I'm speculating that the very last ballots were Federal Overseas Write-in ballots, where you have to write in all office, and if it is not a federal office you have to write-in the office itself.

Persons who had not received a ballot or misplaced it might have used the ballot at the last minute, and it was among last counted. And/or the election officials may have set it aside, because it possible for voters to vote both an official absentee ballot and the write-in ballot.

This might have been especially true in Fairbanks because Wainwright is such a dominant presence.

[jimrtex]

Anchorage absentee complete +309.
HD-37 absentee complete +91
HD-40 absentee complete +304

There are about 1000 Juneau mail ballots uncounted, and an uncertain number of questioned and early votes in HD-40.

Current margin +3856. Juneau will add +100. Final margin 1.12% (50.56% to 49.44%)

I don't think it's Juneau that's left. I think it's Fairbanks region's 1,2,4-6 and 9, plus maybe a little bit of 37. 40 should be done counting.

My guess is 830 votes or so remain - but it's just a guess. Could be up to 1,000.

Yes, I meant Fairbanks.

Their absentee ballots were completed on Wednesday, and I think the early vote and questioned vote were completed as well.

The link to the number of outstanding ballots has also disappeared so I think they are done.

Final margin is 1.10%.

An oddity is that the final ballots were heavily undervoted, except for President. I'm speculating that the very last ballots were Federal Overseas Write-in ballots, where you have to write in all office, and if it is not a federal office you have to write-in the office itself.

Persons who had not received a ballot or misplaced it might have used the ballot at the last minute, and it was among last counted. And/or the election officials may have set it aside, because it possible for voters to vote both an official absentee ballot and the write-in ballot.

This might have been especially true in Fairbanks because Wainwright is such a dominant presence.

Alaska’s federal overseas ballots are in fake HD 99. I don’t think they can vote on state races.

There is a legal distinction between people who have moved overseas permanently (or do not have a definite intent to return to the US). But intent can not be objectively determined, even by an individual themselves.

https://www.elections.alaska.gov/Core/overseaspermanentvoters.php

In any case, Congress has determined such citizens are entitled to vote in federal elections based on their last US domicile. But Alaska doesn't have to let them vote in state and local elections - they may in fact believe that if they ever do return to the US it will not be in Alaska. These persons get federal ballots (District 99) and only include three federal races.

https://www.elections.alaska.gov/Core/SampleBallot_2020_GEN.php

But voters who are temporarily overseas are treated the same as if they were in the US (their domicile has not changed). But the federal government has provided additional requirements such as ballots being mailed 45 days before an election to ensure time for the ballot to travel overseas and back. In general, voted ballots must have a wet signature, and thus can not be faxed or e-mailed, though the unmarked ballot might be (In Texas, voted ballots can be voted by e-mail if used from a .mil address. This might require assent to non-confidentiality. It gets really messy. Counties with lots of military voters (Bexar, Bell, and El Paso) have experimented with special servers that can anonymize and receive ballots - these require a special app that produces an encrypted file, which can be sent back. At other times, overseas voters have been expected to construct an envelope when they received a faxed ballot. The envelope had to be folded and taped together.

In addition, they can send in write-in ballots, which may require writing-in not only candidates but offices as well.

For HD-1 the last batch included 178 presidential votes, 99 senatorial votes, 101 congressional votes, and 89 question two votes. These were not all federal write-in ballots but many were. The write-in ballots might have been set aside to make sure that a regular mail ballot was not also received.

The reason that HD-6 and HD-9, 32, 37, 39, and 40 results were split is because those districts are divided between judicial districts, and have different ballot styles based on judicial races. See sample ballots above. Alaska has relatively few ballot styles because it US representative is elected at large, and house districts nest into senate district. If it weren't for the judicial districts, 41 ballot styles would be needed.

The judicial districts are long established - at one time census results were tabulated by judicial district. They also tend to be more or less permanent, it is messy for venue to change during multi-year litigation. They probably try to avoid have too many house districts crossing judicial districts, but that may hard to accomplish, particularly if keeping bush districts from going into cities.

[jimrtex]

FWIW - this is the first time I've ever seen the HDs split in the txt results file. It will be interesting to see if they're still separated when the results go official.

For the 2018 general election the precinct-level judicial results for HD-6 have the results for JD 3 (Gulf of Alaska) or JD 4 (Yukon and Kuskokwim basins) depending on which judicial district the precinct is in. It is quite likely that no precinct crosses a judicial district boundary.

But for HD-6 absentee ballots, the judicial races are treated as separate contests in the same area even though they weren't.

So for governor there were 14896 ballots, but for JD 3, 2028 ballots and JD 4, 12868 ballots.

14896 = 2028 + 12868.

You may have never looked at judicial races. Retention races are pretty boring unless there is an active campaign against a judge.

Being a modern state, Alaska may have more expedient ways of getting rid of a really corrupt judge.